Code
# Loading packages
library(tidycensus)
library(tidyverse)
library(sf)
library(tigris)
library(patchwork)
library(statebins)
library(maps)
library(tools)
library(giscoR)
library(s2)
library(geosphere)
library(wesanderson)The Shadows of Enumeration:
What Can Learn about Afro-Latinos Using U.S. Census Data
Objective
The goal of this project is to use Census analysis to provide a detailed profile of the Afro-Latino population in the U.S., their geographical spread, socioeconomic characteristics, and how they fare compared to Latinos who identify as White-only and non-Black.
Data
All of the data I used in these analyses come from the U.S. Census Burau, by way of the R package tidycensus.
tidycensus Developed and maintained by Kyle Walker, this package allows for streamlined use of the US Census Buerau’s data API, providing tidy data sets to users without needing to download or clean data sets outside of R studio. tidycensus also enables users to easily use spatial data and geometry, for ease in creating heat maps or population geography charts.
Using tidycensus, I accessed the U.S. Bureau’s following survey objects:
get_acs(), which requests data from the 1-year and 5-year American Community Survey samples. Data are available from the 1-year ACS back to 2005 and the 5-year ACS back to 2005-2009.get_estimates(), an interface to the Population Estimates APIs. These datasets include yearly estimates of population characteristics by state, county, and metropolitan area, along with components of change demographic estimates like births, deaths, and migration rates.get_flows(), an interface to the ACS Migration Flows APIs. Includes information on in- and out-flows from various geographies for the 5-year ACS samples, enabling origin-destination analyses.get_pums(), which accesses data from the ACS Public Use Microdata Sample APIs. These samples include anonymized individual-level records from the ACS organized by household and are highly useful for many different social science analyses.1
5-year ACS Data
- Pros: very precise, up to date
- Cons: does not allow for inclusion of Latinos who select Black and one or more other races (which is likely most Afro-Latinos, given differing ideas on race in Latin America)
Code
# Creating a 5 year ACS data sets
# state hisp pop
state_hisp_pop <- get_acs(
geography = "state",
variables = c(latino_pop = "B03003_003"),
year = 2020,
resolution = "20m",
progress_bar = FALSE
) %>%
select(NAME, estimate) %>%
rename(lat_pop = estimate)
# Black ONLY race
acs_bl <- get_acs(
geography = "state",
variables = c(black_latinos = "B03002_014"),
geometry = TRUE,
year = 2020,
resolution = "20m",
progress_bar = FALSE
) %>%
inner_join(state_hisp_pop) %>%
mutate(
pop_perc = (estimate/lat_pop)*100
) %>%
shift_geometry()
# White ONLY race
acs_wl <- get_acs(
geography = "state",
variables = c(white_latinos = "B03002_013"),
geometry = TRUE,
year = 2020,
resolution = "20m",
progress_bar = FALSE
) %>%
inner_join(state_hisp_pop) %>%
mutate(
pop_perc = (estimate/lat_pop)*100
) %>%
shift_geometry()ACS 2020 Heat Maps
Code
black_only_map <- acs_bl %>%
ggplot(aes(fill = estimate)) +
geom_sf() +
scale_fill_viridis_c(option = "rocket",
limits = c(0, 30000),
breaks = c(0, 15000, 30000),
labels = c("0", "15,000", "30,000 or more"),
oob = scales::squish) +
labs(
fill = "Latinos who identify\nas Black only"
) +
ggtitle("Latino population distribution by Black- or White-only identification") +
theme_void()
white_only_map <- acs_wl %>%
ggplot(aes(fill = estimate)) +
geom_sf() +
scale_fill_viridis_c(option = "mako",
limits = c(0, 650000),
breaks = c(0, 325000, 650000),
labels = c("0", "325,000", "650,000 or more"),
oob = scales::squish) +
labs(
fill = "Latinos who identify\nas white only"
) +
theme_void()
black_only_map / white_only_map 
ACS 2020 state bins
Code
# Population cutting
bl_estimate <- acs_bl$estimate
wl_estimate <- acs_wl$estimate
# Black latino bins
acs_bl <- mutate(
acs_bl,
pop_bins = cut(
ifelse(is.na(bl_estimate), 10000, bl_estimate),
breaks = c(0, 15000, 30000),
right = FALSE
)
)
# Black only Latinos plot (minimizes the Afro-latino pop, avoid including in write-ups)
black_bins <- acs_bl %>%
ggplot() +
statebins::geom_statebins(
mapping = aes(state = NAME,
fill = estimate)
) +
scale_fill_viridis_c(option = "rocket",
limits = c(0, 30000),
breaks = c(0, 15000, 30000),
labels = c("0",
"15K",
"30K+"),
oob = scales::squish) +
ggtitle("Latinos that identify as Black-only") +
theme_statebins()
# White latino bins
acs_wl <- mutate(
acs_wl,
pop_bins = cut(
ifelse(is.na(wl_estimate), 400000, wl_estimate),
breaks = c(0, 325000, 650000),
right = FALSE
)
)
# White only Latino plot
white_bins <- acs_wl %>%
ggplot() +
statebins::geom_statebins(
mapping = aes(state = NAME,
fill = estimate)
) +
scale_fill_viridis_c(option = "mako",
limits = c(0, 650000),
breaks = c(0, 325000, 650000),
labels = c("0",
"325K",
"400K+"),
oob = scales::squish) +
ggtitle("Latinos that identify as White-only") +
theme_statebins()
# Plots in same figure
black_bins 
Code
white_bins
2019 Population Estimates
- Pros: allows for disaggregation of race and latino origin
- Cons: based on Census projections; robust, yet not actual population counts
Estimates heatmaps
Code
# Creating a Census estimates data set
us_estimates <- get_estimates(
geography = "state",
product = "characteristics",
breakdown = c("RACE", "HISP"),
year = 2019,
breakdown_labels = TRUE,
geometry = TRUE,
progress_bar = FALSE
) %>%
shift_geometry() %>%
filter(
RACE == "Black alone or in combination" |
RACE == "White alone" |
RACE == "All races") %>%
filter(HISP == "Hispanic")
tot_hisp_pop_est <- us_estimates %>%
filter(RACE == "All races") %>%
select(GEOID, NAME, value) %>%
rename(total_hisp_pop = value)
# Black alone or any other race
afrolat_state <- us_estimates %>%
filter(RACE == "Black alone or in combination",
HISP == "Hispanic") %>%
st_join(tot_hisp_pop_est) %>%
mutate(
pop_perc = (value/total_hisp_pop)*100
)
# White alone
whitelat_state <- us_estimates %>%
filter(RACE == "White alone",
HISP == "Hispanic") %>%
st_join(tot_hisp_pop_est) %>%
mutate(
pop_perc = (value/total_hisp_pop)*100
)Population proportion state bins
Code
# Subsets for state bins visualization
white_perc_pop_bins <- whitelat_state %>%
ggplot() +
statebins::geom_statebins(
mapping = aes(state = NAME.x,
fill = pop_perc)
) +
scale_fill_viridis_c(option = "mako",
limits = c(70, 90),
breaks = c(50, 75, 90),
labels = c("50","< 75%", "95%"),
oob = scales::squish,
name = "% of State Hispanic/\nLatino Population") +
labs(
title = "White-only Latinos\nas percent of state Latino population"
) +
theme_statebins()
afro_perc_pop_bins <- afrolat_state %>%
ggplot() +
statebins::geom_statebins(
mapping = aes(state = NAME.x,
fill = pop_perc)
) +
scale_fill_viridis_c(option = "rocket",
limits = c(2, 17),
breaks = c(0, 7.5, 17, 20),
labels = c("0","< 7.5%", "20%", "20%"),
oob = scales::squish,
name = "% of State Hispanic/\nLatino Population") +
labs(
title = "Afro-Latinos\nas percent of state Latino population"
) +
theme_statebins()
white_perc_pop_bins
Code
afro_perc_pop_bins
Estimates bubble map
Code
# Loading US map info
map_states <- map_data("state") %>% as_tibble()
map_cty <- map_data("county") %>% as_tibble()
cty_names <- unique(map_cty$subregion)
state_names <- unique(map_states$region)
# Hispanic pop per county
cty_hisp_pop_est <- get_estimates(
geography = "county",
product = "characteristics",
breakdown = c("RACE", "HISP"),
year = 2019,
breakdown_labels = TRUE
) %>%
filter(RACE == "All races") %>%
filter(HISP == "Hispanic") %>%
select(GEOID, NAME, value) %>%
mutate(NAME = tolower(NAME)) %>%
rename(total_hisp_pop = value)
# Polygon data set
afrolat_polygon <- get_estimates(
geography = "county",
product = "characteristics",
breakdown = c("RACE", "HISP"),
year = 2019,
breakdown_labels = TRUE) %>%
filter(RACE == "Black alone or in combination") %>%
filter(HISP == "Hispanic") %>%
mutate(NAME = tolower(NAME)) %>%
mutate(region =
str_extract(NAME,
paste(state_names,
collapse = "|"))) %>%
inner_join(cty_hisp_pop_est) %>%
mutate(pop_perc = (value/total_hisp_pop)*100) %>%
mutate(subregion =
str_extract(NAME,
paste(cty_names,
collapse = "|"))) %>%
inner_join(map_cty, relationship = "many-to-many") %>%
mutate(value = na_if(value, 0)) %>%
drop_na()
# Discarding duplicates for county data
afrolat_polygon2 <- afrolat_polygon %>%
distinct(subregion, .keep_all = TRUE) %>%
drop_na() %>%
filter(value > 29)
# Scaling parameters - making a small quantiles vector
cty_quantiles <- quantile(afrolat_polygon2$value)
# Create the discrete values for population per country using `cut`
levels_breaks <- c(30,
65,
100,
300,
500,
1000,
5000,
10000,
50000,
100000,
300000)
bubble_cut <- cut(afrolat_polygon2$value,
breaks = levels_breaks,
include.lowest = TRUE,
right = FALSE,
ordered_result = TRUE,
dig.lab = 3)
afrolat_polygon2$bubble_cut <- bubble_cut
levels_count <- length(levels(afrolat_polygon2$bubble_cut))
# Create the sizes scale
manual_size <- seq(1, by = 1, length.out = levels_count)
names(manual_size) <- levels(afrolat_polygon2$bubble_cut)
levels_labels <- c("30 - 65",
"65 - 100",
"100 - 300",
"300 - 500",
"500 - 1,000",
"1,000 - 5,000",
"5,000 - 10,000",
"10,000 - 50,000",
"50,000 - 100,000",
"100,000 - 250,000")
# Color scales
levels_colors <- c("[30,65)" = "#fde0dd",
"[65,100)" = "#fcc5c0",
"[100,300)" = "#fa9fb5",
"[300,500)" = "#f768a1",
"[500,1e+03)" = "#dd3497",
"[1e+03,5e+03)" = "#ae017e",
"[5e+03,1e+04)" = "#7a0177",
"[1e+04,5e+04)" = "#49006a",
"[5e+04,1e+05)" = "#51209F",
"[1e+05,3e+05]" = "#221A80"
)
color_scale <- scales::seq_gradient_pal(levels_colors)
# Plotting - need to add long and lat data
base_bubble_plot <- ggplot() +
geom_polygon(data = map_states,
aes(x = long,
y = lat,
group = group),
fill = "gray99",
col = "gray50") +
geom_point(data = afrolat_polygon2,
aes(x = long, y = lat,
size = factor(bubble_cut),
col = factor(bubble_cut)),
alpha = 0.3,
)
final_bubble_plot <- base_bubble_plot +
scale_size_discrete(
name = "Afro-Latinos\nPer County",
labels = levels_labels
) +
scale_color_manual(
name = "Afro-Latinos\nPer County",
labels = levels_labels,
values = levels_colors
) +
labs(
title = "U.S. Afro-Latino Population Per County",
subtitle = "Indiviudals who identified as Hispanic and Black alone or \nin combination with other races (total 2.5 million)",
caption = "Author: Michelle Bueno Vásquez\nData: U.S. Census Bureau 2019 Population Estimates"
) +
theme_void()
final_bubble_plot
PUMS Data
Code
# PUMS Data Wrangling
pums_vars <- c(hisp_org = "HISP",
black = "RACBLK",
mig_country = "MIGSP",
housh_income = "HINCP",
edu_level = "SCHL",
hlth_ins_cov = "HICOV",
citizen = "CIT",
birth_place = "WAOB",
food_stamps = "FS",
house_language = "HHL",
limited_english = "LNGI",
home_lang = "LANP",
internet = "ACCESSINET",
employment = "ESR",
rent_income = "GRPIP"
)
pums_vars_selection <- c("mig_country",
"black",
"hisp_org",
"housh_income",
"edu_level",
"hlth_ins_cov",
"citizen",
"birth_place",
"food_stamps",
"house_language",
"limited_english",
"home_lang",
"internet",
"employment",
"rent_income")
#Loading data for Afro-Latinos
pums_blackhisp_dat <- read_csv("blackhisp_detailed_data") %>%
rename(all_of(pums_vars)) %>%
select(all_of(pums_vars_selection))
# Recoding country codes
pums_afro_tidy <- pums_blackhisp_dat %>%
mutate(hisp_org = case_match(
hisp_org,
"02" ~ "Mexico",
"03" ~ "Puerto Rico",
"04" ~ "Cuba",
"05" ~ "Dominican Republic",
"06" ~ "Costa Rica",
"07" ~ "Guatemala",
"08" ~ "Honduras",
"09" ~ "Nicaragua",
"10" ~ "Panama",
"11" ~ "El Salvador",
"12" ~ "Other Central Am.",
"13" ~ "Argentina",
"14" ~ "Bolivia",
"15" ~ "Chile",
"16" ~ "Colombia",
"17" ~ "Ecuador",
"18" ~ "Paraguay",
"19" ~ "Peru",
"20" ~ "Uruguay",
"21" ~ "Venezuela",
"22" ~ "Other South Am.",
"24" ~ "Spain"
))Afro-Latinos Latin America Countries of Origin
Code
# Names of Lat Am countries
lat_am_countries <- c("Brazil",
"Uruguay",
"Argentina",
"French Guiana",
"Suriname",
"Colombia",
"Venezuela",
"Bolivia",
"Ecuador",
"Chile",
"Paraguay",
"Peru",
"Guyana",
"Panama",
"Costa Rica",
"Nicaragua",
"Honduras",
"El Salvador",
"Belize",
"Guatemala",
"Mexico",
"Trinidad and Tobago",
"Caribe",
"Puerto Rico",
"Dominican Republic",
"Haiti",
"Jamaica",
"Cuba",
"Bahamas",
"Antiles",
"Dominica",
"Saba")
# Creating a map df
map_tib <- map_data(map = "world",
regions = lat_am_countries) %>%
select(-subregion)
# Centroid coordinates for country data
centroids <- map_tib %>%
group_by(region) %>%
group_modify(~ data.frame(centroid(cbind(.x$long, .x$lat)))) %>%
mutate(long = lon) %>%
select(-lon)
# Population count from countries of orgin tibble
tally_tib <- pums_afro_tidy %>%
count(hisp_org) %>%
arrange(desc(n)) %>%
cbind(as.factor(c(23:1))) %>%
rbind(c("Brazil", 0), c("Haiti", 0))
colnames(tally_tib) <- c("hisp_org", "tally", "pop_ranking")
tally_tib <- tally_tib %>%
mutate(tally = as.numeric(tally))
# Adding centroid and tally data to data set
pums_afro_tidy <- pums_afro_tidy %>%
mutate(
region = hisp_org,
) %>%
inner_join(centroids) %>%
inner_join(tally_tib)
# Adding tally data to map tibble
tally_tib <- tally_tib %>%
rename(region = hisp_org)
map_tib <- map_tib %>%
inner_join(tally_tib)Code
lat_am_plot <- ggplot() +
geom_polygon(data = map_tib,
aes(x = long,
y = lat,
group = group,
fill = tally),
col = "gray50",
size = 0.2
) +
scale_fill_viridis_c(option = "inferno",
name = "Population\n(thousands)",
labels = c("0",
"5,000",
"10,000",
"15,000",
"20,000",
"25,000")) +
coord_cartesian(expand = FALSE) +
labs(
title = "Country of origin of Afro-Latinos*",
caption = "*DISCLAIMER: Census data wrongfully excludes Haiti and Brazil from consideration as Latino "
) +
theme_bw()
caribbean_plot <- ggplot() +
geom_polygon(data = map_tib,
aes(x = long,
y = lat,
group = group,
fill = tally),
col = "gray50",
size = 0.2
) +
scale_fill_viridis_c(option = "inferno",
name = "Population\n(thousands)",
labels = c("0",
"5,000",
"10,000",
"15,000",
"20,000",
"25,000")) +
coord_cartesian(xlim = c(-110, -65),
ylim = c(15, 32),
expand = FALSE) +
labs(
title = "Most Afro-Latinos are of Caribbean Ancestry*",
caption = "*DISCLAIMER: Census data wrongfully excludes Haiti and Brazil from consideration as Latino "
) +
theme_bw()
lat_am_plot
Code
caribbean_plot
Code
### Origin Barplot
tally_tib %>%
filter(!region %in% c("Spain", "Haiti", "Brazil", "NA")) %>%
drop_na() %>%
ggplot(aes(x = reorder(region, tally),
y = tally)
) +
geom_col(fill = "cyan4") +
coord_flip() +
labs(x = "Population Count (thousands)",
y = "Country of Origin",
title = "Afro-Latinos by country of origin") +
theme_bw()
Comparisons
Code
# Income data across top 5 national origin groups
income_densities <- pums_afro_tidy %>%
filter(region %in% c("Mexico",
"Puerto Rico",
"Dominican Republic",
"Cuba",
"Panama")) %>%
drop_na() %>%
ggplot(aes(x = housh_income,
fill = region,
color = region)) +
geom_density(aes(y = ..scaled..),
alpha = 0.4) +
scale_fill_brewer(palette = "Set1",
name = "Country of origin") +
scale_color_brewer(palette = "Set1",
name = "Country of origin") +
labs(x = NULL,
y = "Density",
title = "Household income for Afro-Latino by top origin groups") +
scale_x_continuous(labels = scales::dollar,
limits = c(0, 250000)) +
theme_bw() +
theme(
axis.ticks.x = element_blank()
)
income_densities
Code
income_densities +
facet_wrap(~region) +
theme(
axis.ticks.x = element_blank(),
axis.text.x = element_blank()
)
ShinyApp: Dominican population across states
I designed a ShinyApp where users can explore the Dominican population densities in their state of interest using the 2020 ACS data.
The app can be accesed by clicking the section title above. Enjoy!
Footnotes
Source: Analyzing US Census Data: Methods, Maps, and Models in R by Kyle Walker↩︎